Skip to workspace
PixelOkayAudio Studio
Tools
LOCAL-FIRST · NO PAID API

AI Audiobook Creator

Import or record your own narration, master chapter audio and export a complete audiobook locally. Optional AI narration is separate.

● Record / import narration▤ Chapter studio♫ Downloadable WAV◇ No account
02 / WRITE & EDIT

Chapter editor

No edits yet
Chapter audio · original take preservedNo AI

Waveform and editor use converted mono 24 kHz PCM. Effects apply to an output copy; choose the original take to revert.

0 words · 0 characters
♫
No audio selectedGenerate a sample or chapter to listen
A SIMPLE CREATIVE WORKFLOW

From manuscript to audiobook

01

Add your text

Paste or import a UTF-8 TXT/Markdown document. Use headings to identify chapters, then edit names and content.

02

Record or import

Record with your microphone or import audio for each chapter. Optional neural voice generation is separate.

03

Generate & export

Produce on-device speech and download individual WAV chapters, a combined WAV or a ZIP of chapter files. Regenerate only chapters that have changed, and level completed narration when needed.

What “AI” means here

Exportable narration uses the open-weight Kokoro TTS neural model in your browser; it does not send your book to a paid speech service. The explicit Download & Load Model step downloads the runtime, model and voice resources from public hosts. Processing may be slow on low-memory devices, and synthesis quality is optimized for English. Generated speech is not a human recording; check pronunciation and rights to the text before sharing. A browser voice is not used to fake downloadable speech.

Privacy: Manuscript text is processed on your device; an optional locally stored draft uses your browser storage. This tool does not have an upload endpoint. Hosting analytics, CDN requests for model files, and browser caching are separate considerations. Imported and recorded audio is locally decoded to 16-bit mono 24 kHz PCM WAV, not MP3, and does not support DRM or cloning real voices.

LOCAL AI SETUP

Set up AI on this device

First, a model needs to be loaded on your device

Kokoro neural text-to-speech runs locally in a dedicated worker. The first download loads runtime, model weights and voice assets from public hosts. Nothing is downloaded until you select Download & Load Model. Manuscript editing and project exports remain available without AI.

  1. 1. Check device
  2. 2. Download / cache
  3. 3. Prepare narrator
  4. 4. Narrate chapters

Device & model resources

Measurements may be approximate
Device RAM / model memoryChecking device information…

Live available RAM and GPU memory are not exposed by most browsers.

Browser website storageChecking site storage…

Site-wide approximate origin quota, shared with other PixelOkay tools. This is not RAM.

ONE ACTION · ALL RECOGNIZED AI MODEL DOWNLOADS

Free AI model storage across PixelOkay

Scan this website origin for WebLLM and Transformers.js model caches, plus recognizable model files in other cache buckets. This includes caches from other PixelOkay tools, not only this tool. No site data or browser databases are cleared automatically.

Run a scan to inspect recognized model caches before optional cleanup.

Unknown IndexedDB, OPFS and browser-managed models remain untouched. Other AI tabs can hold models in use; close them before cleanup.

Having trouble downloading or loading AI?

Refresh storage/cache, retry setup, or inspect recognized same-origin model caches. Storage deletion is optional and will not fix CDN, CORS, WASM or device-memory failures.

Clearing all website data can erase saved work across PixelOkay; this tool never performs that action.

Kokoro ONNX q8 / WASM. Loading status unknown until you explicitly start setup.

Technical details and safe recovery

kokoro-js 1.2.1 · Kokoro-82M-v1.0-ONNX · q8 · WASM; manuscript text excluded from diagnostics.

Do not clear all browser website data without exporting your projects and files. This tool does not access or remove them.

Kokoro creates 24 kHz WAV narration locally; its CDN/model downloads use external hosts. Cache inspection covers only verified Cache Storage entries. Kokoro files in browser-managed HTTP, IndexedDB or OPFS storage cannot safely be inspected or erased here. Cached Kokoro status is unknown until the runtime actually loads. No cross-runtime binary reuse is assumed.
CROSS-TOOL MODEL CACHE CLEANUP

Delete recognized AI model downloads?

This may make other PixelOkay AI tools redownload shared model weights. Saved projects, imported files, and browser databases are not targeted. Unknown caches are skipped. Close other AI tabs first.

Using AI Audiobook Creator

Import or write a manuscript, arrange chapters and set voice/speed per chapter. Generate requested narration, review current outputs and optionally level chapter audio. Save the text project and generated audio separately before refreshing.

No AI is required for recording or importing chapter audio. Kokoro downloads model/runtime assets only after the optional AI setup action and explicit Download & Load Model consent and can be slow or memory-intensive. English voices and generated pronunciation are not guaranteed for every text. WAV/ZIP are supported; MP3/M4B are not. Chapter/text/segment/PCM budgets apply. Text drafts may be stored locally; audio is included only in an explicitly downloaded portable-project ZIP, not plain JSON. Editing chapter settings makes old audio stale.

References: Kokoro ONNX model

Turn a reviewed manuscript into chapter audio

Organize text into chapters, choose supported voice settings, and request optional Kokoro narration. Review generated chunks and chapter boundaries before exporting supported WAV or ZIP output; a text draft or device preview is not itself exported narration.

Example workflow

  1. Prepare a short manuscript with clear chapter breaks and check names, punctuation and pronunciation-sensitive words.
  2. Generate a chapter sample, listen for pronunciation and pacing, then continue in bounded chunks while checking completion status.
  3. Review each chapter and export generated WAV or ZIP; retain manuscript and project settings separately.

Authored example: Divide a short fable into two chapters and test a proper name in the opening paragraph. Listen to the generated sample before processing the rest, and revise text or voice settings if the name is unclear.

Limits and model handling: Kokoro and its ONNX model load on Generate or Preview and may need substantial memory; generated pronunciation varies. WAV and ZIP are supported, while MP3 and M4B are not. Text drafts may persist locally, but narration audio is not persisted or included in project JSON. Editing chapter settings can make existing audio stale.

AI Audiobook Creator questions

Can I export an MP3 or M4B audiobook?

No. The supported audio package uses WAV and ZIP. Convert separately if your downstream service requires another format.

Does project JSON contain generated narration?

No. Plain JSON stores manuscript and settings only. Use Portable project + audio ZIP to reopen your narration offline.

Related tools